NVDA price vs trailing EPS — did the multiple compress over the last 3 years?
NVDA’s parabolic climb looks, counterintuitively, like an earnings story: over the roughly three-year window analyzed the share price rose about +372.15% while trailing‑12M EPS exploded roughly +3,293.87%, driving the P/E down from 219.65 to 30.56 (an −86.09% change). Put simply, earnings growth materially outpaced price appreciation, so valuation compressed rather than stretched.
I built a daily NVDA close series across 752 trading days, constructed TTM EPS from quarterly diluted EPS with a 60‑day reporting lag, forward‑filled those values, and computed price/TTM. A linear fit to log(P/E) shows a decline of about −0.4800 per year (p≈5.2e‑212). The charts, full statistics, and robustness checks follow below.
For NVDA over the past ~3 years, has the parabolic run actually made the stock cheaper on earnings — did trailing EPS grow even faster than the share price, compressing the P/E multiple rather than inflating it? Thesis: trailing 12-month EPS outpaced price over the window, so NVDA's P/E actually contracted despite the multi-hundred-percent rally, meaning the move is an earnings story, not a stretched-multiple bubble.
How this was measured
Built a daily NVDA close series over ~3 years. Constructed trailing-12M EPS (TTM) from quarterly eps_diluted by summing the last four quarters and applying a 60-day post-fiscal-period reporting lag as the first availability date (effective_date). Forward-filled TTM onto trading days, computed P/E = price/TTM EPS, and compared start vs end levels and CAGRs. Fitted a linear trend to log(P/E) vs time (years) to test for systematic contraction or expansion.
The key numbers
Reading the numbers
Across the ~3-year window NVDA's price rose from its start to a rebased 472.1453 while TTM EPS exploded to a rebased 3393.8669; as a result the trailing P/E fell from 219.6481 to 30.5568 (about an 86% contraction). The headline slopes show log P/E trending down (-0.4800/yr, p≈5.21e-212).
The charts
This chart rebases both series to 100 at the start so you can compare growth rates directly: Price ends at 472.1453 while TTM EPS ends at 3393.8669. The striking detail is how far above the price line the EPS line finishes — EPS grew roughly an order of magnitude faster than price over the window. That gap is the mechanical reason trailing EPS could outpace price and compress the P/E multiple.
The P/E multiple falls from 219.6481 at the start to 30.5568 at the end, with a series mean of 79.0922 and a recent value very close to the window minimum (min 29.8815). The most important thing to look at is that ending P/E sits well below the long-run average and near the low, indicating clear multiple compression rather than expansion. The log P/E slope of -0.4799730368 per year with p≈5.21e-212 confirms this downward trend is statistically detectable, which supports the thesis that earnings growth, not multiple inflation, drove the move.
Start vs End snapshot
| point | date | price_usd | ttm_eps_usd | pe |
|---|---|---|---|---|
| Start | 2023-06-30 | 42.26 | 0.1924 | 219.65 |
| End | 2026-06-30 | 199.53 | 6.5298 | 30.56 |
The takeaway
Yes — over the ~3-year window NVDA’s rally was accompanied by far faster EPS growth, so the P/E multiple collapsed rather than stretched. TTM EPS rose about +3,293.87% while price rose about +372.15%, driving P/E from 219.65 down to 30.56 (an -86.09% total drop). The fitted trend shows log(P/E) falling at roughly -0.4800 per year and the slope’s p-value is ~5.2e-212, based on 752 trading days — in plain terms, this is a real, very strong signal, not a coin flip. Practically: the move looks like an earnings story (massive EPS expansion) that compressed valuation, not a pure multiple blowoff; the multiple contraction is large and statistically robust.
The fine print
- TTM EPS is constructed from quarterly diluted EPS and only becomes available 60 days after each fiscal period; release-timestamp precision could change short-term alignment.
- TTM EPS updates quarterly while price is daily, so intra-quarter P/E volatility is driven entirely by price moves.
- The very high start P/E was driven by a very small starting EPS; trimming or longer history could moderate absolute multiples.
- This covers ~36 months only; a longer sample might show different longer-term multiple regimes.